In examining commercial typefaces for the release of my own in-progress font, I've been very surprised to find that a number of TrueType fonts, such as Trade Gothic LT Std Bold and Folio Bk BT Book don't include diacritical variations of the basic alphabet in their kerning pairs. In other words, "AV" and "LY" are kerned fine, but "ÄV" and "LŸ" are un-kerned entirely.
My understanding is that TrueType doesn't support kerning classes, which understandably makes the total number of kerning permutations for all such variants pretty huge, but that raises two obvious questions:
1) Difficulty aside, how does that matter? Typographers using diacriticals don't have less need for kerning simply because of the limitations of a font format. I can't imagine the world was waiting for OpenType to emerge for years before it'd finally have complete kerning capabilities.
2) Surely tools must exist (or did exist) for designing a font with a class-based system that then simply export a traditional, flat list of TrueType kerning pairs upon completion. The class-based approach, as I understand it, really only benefits the developer anyway. It's not as if end users of a font are directly making use of class-based representation of kerning pairs (again, as I understand it).
So what am I missing here? As an American English speaker I rarely have use for diacriticals, but even I find it hard to believe that this sizeable portion of the worldwide Latin alphabet is routinely left un-kerned by professional type designers when targeting the TrueType format. Furthermore, I was planning targeting TrueType myself for simplicity's sake (as my typeface has no real need for OpenType features), but if there's some technical reason why I can't kern the entire set of diacriticals as well, that's a deal breaker to be sure.